Abstract: On the basis of literature, the paper sorts out factors influencing sense of security in streets. It also tests western street security theories, such as the Street Eye theory, with the Shanghai samples. It invites 30 students and 30 citizens to evaluate 300 Baidu street view pictures of Shanghai communities in five different development periods in terms of sense of security. The results showed that the green vision rate, management level and number of lanes leave significant impacts on sense of security, while different impact on perceived safety. It sets up security regression models for single- and double-lane and multi-lane street space, finding that the green vision rate (single- and double-lane correlation coefficient R=0.728, p<0.01; multi-lane correlation coefficient R=0.471, p<0.01), management level (single- and double-lane correlation coefficient R=0.766, p<0.01; multi-lane correlation coefficient R=0.450, p<0.01), and number of lanes(correlation coefficient R=0.502, p<0.01) have a significant positive effect on the sense of security. Interface transparency (single- and double-lane correlation coefficient R=0.222, p=0.010), independent bicycle lane (correlation coefficient R=0.309, p<0.01) and design aesthetics factors (correlation coefficient R=0.432, p<0.01) have a positive impact in the single and double lane space. Scooters and bicycles (single- and double-lane correlation coefficient R=-0.327, p<0.01; multi-lane correlation coefficient R=-0.281, p<0.01)have a negative impact on safety perception evaluation. Motor vehicles (single- and double-lane correlation coefficient R=0.251, p<0.01; multi-lane correlation coefficient R=-0.327, p<0.01) present opposite effect in single and double lanes, and multi-lane space.